Author page alternate template (topics-focused) (#3585)

Added a route to get an ArchieML output of all work produced by an author from both gdoc articles and wordpress pages.

- Example route: http://staging-site-author-page-alt/admin/api/all-work?author=Esteban%20Ortiz-Ospina 
- Example page: http://staging-site-author-page-alt/admin/gdocs/1pU8nVBchpI_LYZmNOo4Dag9lHvQkzG-r8JD_57_-1rw/preview

### Why make this change?

To provide a way to manually populate the secondary section of research and writing blocks on author pages with topics rather than articles. This is relevant for a handful of authors who mostly work on topic pages, as well as data managers.

![Screenshot 2024-05-07 at 14.31.06.png](https://graphite-user-uploaded-assets-prod.s3.amazonaws.com/0SFFiIjKuUK6UPYHVe6u/16ed5a76-2cef-4b0e-aa0e-8dcbea95ac73.png)

adminSiteServer/apiRouter.ts

@@ -148,6 +148,7 @@ import { match } from "ts-pattern"
 import { GdocDataInsight } from "../db/model/Gdoc/GdocDataInsight.js"
 import { GdocHomepage } from "../db/model/Gdoc/GdocHomepage.js"
 import { GdocAuthor } from "../db/model/Gdoc/GdocAuthor.js"
+import path from "path"
 
 const apiRouter = new FunctionalRouter()
 
@@ -2543,4 +2544,114 @@ deleteRouteWithRWTransaction(
     }
 )
 
+// Get an ArchieML output of all the work produced by an author. This includes
+// gdoc articles, gdoc modular/linear topic pages and wordpress modular topic
+// pages. Data insights are excluded. This is used to manually populate the
+// [.secondary] section of the {.research-and-writing} block of author pages
+// using the alternate template, which highlights topics rather than articles.
+getRouteWithROTransaction(apiRouter, "/all-work", async (req, res, trx) => {
+    type WordpressPageRecord = {
+        isWordpressPage: number
+    } & Record<
+        "slug" | "title" | "subtitle" | "thumbnail" | "authors" | "publishedAt",
+        string
+    >
+    type GdocRecord = Pick<DbRawPostGdoc, "id" | "publishedAt">
+
+    const author = req.query.author || "Max Roser"
+    const gdocs = await db.knexRaw<GdocRecord>(
+        trx,
+        `-- sql
+            SELECT id, publishedAt
+            FROM posts_gdocs
+            WHERE JSON_CONTAINS(content->'$.authors', '"${author}"')
+            AND type NOT IN ("data-insight", "fragment")
+            AND published = 1
+    `
+    )
+
+    // type: page
+    const wpModularTopicPages = await db.knexRaw<WordpressPageRecord>(
+        trx,
+        `-- sql
+        SELECT
+            wpApiSnapshot->>"$.slug" as slug,
+            wpApiSnapshot->>"$.title.rendered" as title,
+            wpApiSnapshot->>"$.excerpt.rendered" as subtitle,
+            TRUE as isWordpressPage,
+            wpApiSnapshot->>"$.authors_name" as authors,
+            wpApiSnapshot->>"$.featured_media_paths.medium_large" as thumbnail,
+            wpApiSnapshot->>"$.date" as publishedAt
+        FROM posts p
+        WHERE wpApiSnapshot->>"$.content" LIKE '%topic-page%'
+        AND JSON_CONTAINS(wpApiSnapshot->'$.authors_name', '"${author}"')
+        AND wpApiSnapshot->>"$.status" = 'publish'
+        AND NOT EXISTS (
+            SELECT 1 FROM posts_gdocs pg
+            WHERE pg.slug = p.slug
+            AND pg.content->>'$.type' LIKE '%topic-page'
+        )
+        `
+    )
+
+    const isWordpressPage = (
+        post: WordpressPageRecord | GdocRecord
+    ): post is WordpressPageRecord =>
+        (post as WordpressPageRecord).isWordpressPage === 1
+
+    function* generateProperty(key: string, value: string) {
+        yield `${key}: ${value}\n`
+    }
+
+    const sortByDateDesc = (
+        a: GdocRecord | WordpressPageRecord,
+        b: GdocRecord | WordpressPageRecord
+    ): number => {
+        if (!a.publishedAt || !b.publishedAt) return 0
+        return (
+            new Date(b.publishedAt).getTime() -
+            new Date(a.publishedAt).getTime()
+        )
+    }
+
+    function* generateAllWorkArchieMl() {
+        for (const post of [...gdocs, ...wpModularTopicPages].sort(
+            sortByDateDesc
+        )) {
+            if (isWordpressPage(post)) {
+                yield* generateProperty(
+                    "url",
+                    `https://ourworldindata.org/${post.slug}`
+                )
+                yield* generateProperty("title", post.title)
+                yield* generateProperty("subtitle", post.subtitle)
+                yield* generateProperty(
+                    "authors",
+                    JSON.parse(post.authors).join(", ")
+                )
+                const parsedPath = path.parse(post.thumbnail)
+                yield* generateProperty(
+                    "filename",
+                    // /app/uploads/2021/09/reducing-fertilizer-768x301.png -> reducing-fertilizer.png
+                    path.format({
+                        name: parsedPath.name.replace(/-\d+x\d+$/, ""),
+                        ext: parsedPath.ext,
+                    })
+                )
+                yield "\n"
+            } else {
+                // this is a gdoc
+                yield* generateProperty(
+                    "url",
+                    `https://docs.google.com/document/d/${post.id}/edit`
+                )
+                yield "\n"
+            }
+        }
+    }
+
+    res.type("text/plain")
+    res.send([...generateAllWorkArchieMl()].join(""))
+})
+
 export { apiRouter }

db/model/Gdoc/rawToEnriched.ts

@@ -1636,40 +1636,6 @@ function parseExpandableParagraph(
 function parseResearchAndWritingBlock(
     raw: RawBlockResearchAndWriting
 ): EnrichedBlockResearchAndWriting {
-    const createError = (
-        error: ParseError,
-        heading = "",
-        hideAuthors = false,
-        primary = [
-            {
-                value: { url: "" },
-            },
-        ],
-        secondary = [
-            {
-                value: { url: "" },
-            },
-        ],
-        more: EnrichedBlockResearchAndWritingRow = {
-            heading: "",
-            articles: [],
-        },
-        latest: EnrichedBlockResearchAndWritingRow = {
-            heading: "",
-            articles: [],
-        },
-        rows: EnrichedBlockResearchAndWritingRow[] = []
-    ): EnrichedBlockResearchAndWriting => ({
-        type: "research-and-writing",
-        heading,
-        "hide-authors": hideAuthors,
-        primary,
-        secondary,
-        more,
-        latest,
-        rows,
-        parseErrors: [error],
-    })
     const parseErrors: ParseError[] = []
 
     function enrichLink(
@@ -1729,12 +1695,10 @@ function parseResearchAndWritingBlock(
         })
     }
 
-    if (!raw.value.primary)
-        return createError({ message: "Missing primary link" })
     const primary: EnrichedBlockResearchAndWritingLink[] = []
     if (isArray(raw.value.primary)) {
         primary.push(...raw.value.primary.map((link) => enrichLink(link)))
-    } else {
+    } else if (raw.value.primary) {
         primary.push(enrichLink(raw.value.primary))
     }
 

packages/@ourworldindata/types/src/gdocTypes/ArchieMlComponents.ts

@@ -852,8 +852,8 @@ export enum SocialLinkType {
 }
 
 export type RawSocialLink = {
-    text: string
-    url: string
+    text?: string
+    url?: string
     type?: SocialLinkType
 }
 
@@ -862,7 +862,11 @@ export type RawBlockSocials = {
     value: RawSocialLink[] | ArchieMLUnexpectedNonObjectValue
 }
 
-export type EnrichedSocialLink = RawSocialLink
+export type EnrichedSocialLink = {
+    text: string
+    url: string
+    type?: SocialLinkType
+}
 
 export type EnrichedBlockSocials = {
     type: "socials"

packages/@ourworldindata/utils/src/image.ts

@@ -48,7 +48,7 @@ export function getFilenameWithoutExtension(
 export function getFilenameExtension(
     filename: ImageMetadata["filename"]
 ): string {
-    return filename.slice(filename.indexOf(".") + 1)
+    return filename.slice(filename.lastIndexOf(".") + 1)
 }
 
 export function getFilenameAsPng(filename: ImageMetadata["filename"]): string {